Rajiv Ramaswami: Yes. So, George, I can’t say that we have seen increased repatriation as well. We have certainly seen instances of repatriation. What I would say is interesting is that we have also seen stuff in the other direction, which is people have committed to some of these large public cloud spend commitments. And what they are saying, they are not able to get their applications really to public cloud as quickly as they would like to see and so they are now stuck with saying, I have to spend this much money in the public cloud because I committed to it already. And so we have seen that come to us because now they are looking at this saying, we are available through the marketplace of Azure and AWS. And we have seen people purchasing our software through the Azure and AWS marketplace, which help them retire some of their commitment in public spend.
So, yes, I think we see the world being hybrid, some work loads are being repatriated from the public cloud. I wouldn’t say it’s necessarily a wholesale trend, we see that in spots versus a systemic trend yet at this point. But the world is very much hybrid, people are being much more conscious about how much to put in the public cloud in the first place.
George Wang: Okay. Great. Just a quick follow-up, if I can, we simplified portfolio kind of platform approach. Can you talk about kind of the tax rate to Nutanix Cloud Platform, Cloud Manager and the kind of maybe high-level commentary on cross-sell and up-sell?
Rajiv Ramaswami: Yes. So, I think two questions there. On the attach rate of the portfolio, for sure, the biggest change that we have seen since we launched our revised portfolio with much higher attach for cloud management along with our cloud infrastructure. In fact, that is the easiest attach sale, really, right. Everybody who built the cloud also wants to operate the cloud. So, naturally attaching cloud operations, cloud management to our infrastructure. That’s the best we have seen. We have also seen, of course, NC2, which is our public cloud extension is a natural extension, natural attach and it’s really – it ties together with our cloud platform very nicely. So, that’s another attach that is still relatively small for us, but growing every quarter in terms of customers, number of customers and customer accounts.
Unified storage is the other element of the portfolio that we see reasonable attach with as well. The one that we have to actually really work with a specialist team to sell is our Nutanix database portfolio because that’s largely sold a level up in the stack to the people who are either managing databases or at developers that requires more specialist skills, and that’s much more of a specialist.
George Wang: Okay. Great. Thanks. Congrats again.
Rajiv Ramaswami: Thank you.
Operator: Thank you. And one moment for our next question. And our next question comes from Mehdi Hosseini from SIG. Your line is now open.
Mehdi Hosseini: Yes. Thanks for taking my question. I want to go back to your fiscal year ‘24 guide. If I just take the midpoint, it seems like OpEx would need to grow by 7% and I understand you honestly, it’s coming, but would the collaboration or the deal with Cisco actually help? Is some of these OpEx growth assumptions built into the model, or does that order be bake into account, or do that account for any cost savings as Cisco scales your product…?
Rukmini Sivaraman: Hi Mehdi. Thank you for that question. So, let me first address the OpEx point, then we will talk about any potential impact from Cisco. So, if you look at our operating expense profile over the last several years, since about 3 years ago, it was about $1.5 billion and last year it was $1.4 billion, right. It was actually, we have kind of really kept it sort of flat to down over the last 3 years, while of course continuing to grow ACV billings, the ARR revenue and so on since over that time. So, we have been really sort of cautious about OpEx and then focusing on improved efficiency and productivity. Where we are now is that we are making some very prudent, I would say and targeted investments in our go-to-market and in our innovation engine.
As we talked about, our market, we believe is large and growing in order to go and capitalize on that. So, we are making some targeted investments on that front. Now, on the Cisco piece, I think what we expect the Cisco partnership for us to do is to continue to help sort of our productivity go up and really be additive to our overall top line, right. So, that’s how we should think about operating expenses relative to last year, but also relative to any potential impact on Cisco.
Mehdi Hosseini: Okay. And one follow-up for Rajiv, I just want to go back to your prepared remarks. And I am just trying to better understand, I kind of agree with you that security, privacy, compliance would actually bode well for enterprises to scale AI. But as you think about the train models moving from a cloud infrastructure into on-prem, would that actually require your existing or prospective customer with additional one-time CapEx investment or no?
Rajiv Ramaswami: Yes. So, that depends on what they have – I mean obviously, this AI requires compute resources, right. It requires several platforms with GPUs attached to it. So clearly, these are new workloads for the customer, they are going to have to buy new hardware, right, so – and put our software on it. What they would do, just to clarify what they would do, the models likely are combinational models that are trained on publicly available data in the public cloud. But then when they bring it on-prem, they are going to fine tune this, and potentially make it more compact with their own specific data. A good example just to see how this works, maybe it’s – take support as an example, even after the support, in terms of support, we have a lot of internal documentation that we don’t put in the public cloud, internal design documents as well.
So, imagine using a generative AI back end to be able to go through all of that and quickly try and arrive at root cause or try and arrive at answers to customer support questions. So, we want to run that on-prem or in a secure location where our data is stored. And so we will typically run that on, again, standard servers with GPU accelerators, running DJI models with our stack.
Mehdi Hosseini: Thank you. Is that where Cisco could come in and actually become a partner to facilitate that additional investment required?
